Spatially disaggregated working-age population and job estimates for seven Polish cities with a 2025 allocation horizon
Description
This dataset provides spatially disaggregated estimates of working-age population at residential locations and jobs at workplace locations for seven Polish cities: Gdańsk, Kraków, Łódź, Poznań, Szczecin, Warsaw and Wrocław. The estimates were produced using a nominal 100 × 100 m allocation grid by integrating counts from municipal transport studies and models with built-up volume data from the Global Human Settlement Layer (GHSL), release R2023A. Source counts were supplied as attributes of transport-zone polygons. We defined working-age population as women aged 18–59 years and men aged 18–64 years. We calculated residential built-up volume by subtracting non-residential volume from total built-up volume. We then distributed the working-age population within each transport zone in proportion to residential volume, and jobs in proportion to non-residential volume. Where grid cells crossed transport-zone boundaries, their volumes were apportioned according to the area of intersection. We then aggregated allocated fragments by their original grid-cell identifiers. We applied no additional spatial masks or land-use exclusions. For all cities except Poznań, we used the GHSL 2025 built-up volume epoch for allocation. In Poznań, we initially allocated source counts from 2019 using 2020 built-up volumes and then updated them using changes in residential and non-residential volume between 2020 and 2025. The update retained the baseline zone-specific ratios of population or jobs to built-up volume, without renormalising the resulting estimates to the original zone totals. The dataset has a common 2025 allocation horizon, but its source counts do not represent a uniform observation year. In particular, the Łódź inputs refer to 2023, while the Szczecin inputs represent a 2025 forecast from the 2016 traffic study. The GHSL 2025 epoch is also modelled rather than directly observed. The resulting values should therefore be interpreted as spatial allocations and model-based estimates. The two GeoPackage files contain 138,601 working-age population polygons and 50,631 job polygons, representing approximately 3.62 million working-age residents and 3.25 million jobs. The accompanying manuscript assessed their spatial distributions by comparing them with function-specific building floor-area proxies derived from BDOT10k and through spatial autocorrelation analysis. These assessments describe consistency with building capacity and spatial structure, rather than direct accuracy against observed population or employment counts. The dataset supports research on urban accessibility, transport planning, the spatial relationship between residential locations and employment opportunities, and land-use–transport interactions. It does not contain individual records or origin–destination commuting flows.
Files
Steps to reproduce
Step 1. Prepare transport-zone data Obtain transport-zone polygons with working-age population and job counts for each city. Record source years, model versions and forecast scenarios. Define working-age population as women aged 18–59 and men aged 18–64. Use 2019 counts for Poznań, 2023 inputs for Łódź and the 2025 forecast from Szczecin’s 2016 traffic study. Step 2. Prepare GHSL volumes Obtain GHS-BUILT-V R2023A total and non-residential built-up volume rasters at 100 m resolution. Use the 2025 epoch for six cities and the 2020 and 2025 epochs for Poznań. Calculate residential volume by subtracting non-residential volume from total volume. Step 3. Intersect grids and zones Convert the GHSL grid to vector polygons and harmonise coordinate systems with the transport zones. Intersect cells with zones. Apportion each cell’s residential and non-residential volumes using the fraction of its area within each zone. Step 4. Allocate population and jobs Within each zone, multiply each fragment’s share of total residential volume by the source working-age population count. Independently, multiply its share of non-residential volume by the source job count. Use 2025 volumes except in Poznań, where the baseline uses 2020 volumes. Allocation preserves source-zone totals before temporal adjustment, subject to complete coverage and numerical precision. Step 5. Update Poznań to 2025 Calculate residential and non-residential volume changes between 2020 and 2025. Divide each zone’s source population or job count by its corresponding total 2020 volume to obtain a baseline count-to-volume coefficient. Multiply each fragment’s volume change by this coefficient and add the result to its baseline allocation. Apply the same coefficient to locations with zero baseline volume and positive 2025 volume. Update population and jobs separately, without renormalising totals. Step 6. Aggregate fragments Aggregate fragments by original grid-cell identifiers, summing contributions from intersecting zones. Retain city labels and fractional estimates. Released layers contain polygons with positive allocated values. Step 7. Assess spatial consistency Compare estimates with BDOT10k building floor-area proxies derived from footprints and storey counts, selected by building function. Apportion floor area to output polygons by footprint overlap. Assess associations using Spearman correlations and spatial dependence using Global Moran’s I. BDOT10k supports validation and does not mask the allocation; these checks do not directly establish count accuracy. Step 8. Export datasets Export population and jobs as separate GeoPackages, retaining geometries, identifiers, city names and allocation fields. Preserve fractional values during storage and aggregation. Workers_25 represents working-age population, not employed residents; Jobs_25 represents allocated jobs.